Field of the Invention
The present invention relates to image processing of correcting the influence of a defective pixel of an imaging device, which is included in an image.
Description of the Related Art
An imaging device formed by two-dimensionally arranging solid-state image sensors each made of amorphous silicon or single-crystal silicon has been prevalently put into practical use. Such imaging device is widely used not only for a digital camera which captures a visible light image but also for a medical apparatus which captures a radiographic image.
An image captured using the above-described imaging device generally includes a pixel influenced by fixed pattern noise caused by a dark current and a defective pixel caused by a small flaw specific to the image sensor. To create a more preferable image, it is necessary to remove the fixed pattern noise in the captured image and correct the influence of the defective pixel. As correction processing, there are known dark correction processing of removing fixed pattern noise by subtracting a dark image obtained by performing imaging in an unexposed state from an image captured in an exposed state, defect correction processing of interpolating the influence of a defective pixel from surrounding pixels, and the like.
As a technique of correcting the influence of a defective pixel in a captured image, it is common practice to detect in advance a defective pixel at the time of manufacturing an imaging device and store the address of the defective pixel in a memory. In the imaging device, however, it is known that a phenomenon occurs in which a pixel behaving like a defect changes depending on imaging conditions. Correcting, as a defective pixel, a pixel which is not defective increases the processing time and degrades the image quality. Displaying a defective pixel without correction fatally influences the image quality. Therefore, which pixel is selected as a defective pixel according to a situation and how to properly perform defect correction are important issues.
To cope with such situation, there has been proposed a technique of switching a defect correction method in accordance with specific imaging conditions. For example, Japanese Patent Laid-Open No. 2004-222143 (literature 1) proposes a method of switching a defect correction method in accordance with an imaging mode, shutter speed, sensitivity, and luminance at the time of imaging. Also, Japanese Patent No. 4819479 (literature 2) proposes a method of switching a defect correction method in accordance with the accumulation time and gain setting of an image sensor.
However, in a given type of imaging device, a phenomenon occurs in which a pixel behaving like a defect changes in accordance with the accumulation state of electric charges by a dark current generated in an image sensor. For example, a pixel does not exhibit a defect-like behavior when imaging is performed and dark correction is performed immediately after a dark image is obtained, and the same pixel may behave like a defective pixel when imaging is performed and dark correction is performed when a given time has elapsed after obtaining the dark image.
To perform appropriate defect correction for such imaging device, it is necessary to dynamically switch a defect correction method in accordance with the accumulation state of electric charges by a dark current. The techniques described in literatures 1 and 2 cannot solve the above problem since a defect correction method is switched in accordance with imaging conditions.